Joint torque seems insufficient

[johnrowlay]

I have loaded the Sawyer robot arm into a new scene, set all joints to force-torque mode, with very large maximum torques (10^11) and "lock motor when target velocity is zero" checked. If I start running the simulation, as expected the robot does not move.

If I then set the "Target velocity" to 1 deg/s for joint 1, then the robot starts moving that joint very slowly. Again, as expected.

However, if instead of joint 1, I set joint 2 to 1 deg/s, something strange happens. The robot starts moving joint 2 very slowly at first, but soon, the speed builds up and it starts moving very quickly, much faster than 1 deg/s. It is as if the robot is falling under its own weight; whilst joint 1 does not have very large moments to support, joint 2 has very large moments since the entire length of the robot is pushing down on joint 2.

So, why is this happening? Why can joint 2 not achieve the velocity I have set? Is it because it simply does not have enough torque to apply? If so, how can I increase the torque? I cannot increase it beyond 10^11 in the GUI.

Thanks.

[RobAtLab]

This is probaboly a physics engine problem due to having too low a mass and/or moment of inertia on some of your parts, the physics engine then doesn't properly makew these move when you try to turn them slowly, however high your torque may be. You might also find joints floating off axis as another symptom of physics engine troubles. Try increasing the mass and (especially) the moment of inertia of the child objects of joint 2. With this done you'll be able to reduce the torque to a slightly less gargantuan value and still have things work.